Date of Award




Document Type


Degree Name

Doctor of Philosophy (PhD)


Department of Computer Science

Content Description

1 online resource (xiv, 137 pages) : illustrations (some color)

Dissertation/Thesis Chair

Shaghayegh Sahebi

Committee Members

Mei-Hwa F. Chen, Petko Bogdanov, Ming-Ching Chang


Stochastic models

Subject Categories

Computer Sciences


A temporal point process can be viewed as a collection of random points falling in the space of time, which is a special type of stochastic processes that is used to model complex event sequences in continuous time.As event data has become more widely available, temporal point process models (TPPs), i.e. techniques for modeling temporal point processes, have been used to solve a wide range of real-world problems, in domains such as e-commerce, online education, and social media. Motivated by the limitations in TPP literature, this dissertation aims to explore and study the following research questions: 1) Focusing on the limitation of modeling temporal point processes independent and identically distributed (i.i.d), we want to study the question of how to model the relationship among temporal point processes. 2) To address the problem of ineffective representation of time dependencies, the second research question we want to answer is how to effectively represent external stimuli in temporal point process modeling. 3) To tackle the limitation and the scarcity of temporal point process modeling with markers, we want to answer the research question of how to efficiently model markers in temporal point processes while capturing the interrelationship between markers, activity timings and types. 4) Motivated by the difficulty of causality modeling in continuous time, we want to explore how to identify Granger causality in temporal point processes with complex data dynamics. Answering these questions will improve not just the field of temporal point process modeling in terms of model capacity and interpretation, but also potentially a wide range of disciplines with substantial societal impacts, such as in fiance, health, recommendation, and education. To answer these research questions, we develop a series of novel TPPs with high capacity in event prediction as well as meaningful interpretations of data dynamics for a variety of real-world problems.